Health care utilization and HIV clinical outcomes following Affordable Care Act implementation in California: A longitudinal study


 Background: The Affordable Care Act (ACA) increased insurance coverage for people with HIV (PWH) in the United States, yet post-ACA health care utilization and HIV outcomes have rarely been examined. Methods: Among PWH newly enrolled in an integrated health care system (Kaiser Permanente Northern California) in 2014 (N=880), we examined use of health care and modeled associations between enrollment mechanisms (enrolled in a Qualified Health Plan through the California Exchange vs. other sources), deductibles (none, $1-$999 and >= $1000), receipt of benefits from the California AIDS Drug Assistance Program (ADAP), demographic factors, and three-year patterns of health service utilization (primary care, psychiatry, substance treatment, emergency, inpatient) and HIV outcomes (CD4 counts, viral suppression). Results: Health care use was greatest immediately after enrollment and decreased over three years. Those with high deductibles were less likely to use primary care (OR=0.64, 95% CI=0.49 - 0.83, p < 0.001) or psychiatry OR=0.58, 95% CI=0.36, 0.93, p=0.02) than those with low or no deductibles. Enrollment via the Exchange was associated with fewer psychiatry visits (rate ratio [RR]=0.38, 95% CI=0.17 - 0.83; p=0.02), but ADAP was positively associated with number of psychiatry visits (RR=2.22, 95% CI=1.22 – 4.05 p=0.01). There was no association between enrollment mechanism or deductibles and HIV outcomes, but ADAP enrollment was associated with viral suppression (OR=2.23, 95% CI=1.35 - 3.68, p<0.001). Non-white PWH were less likely to be virally suppressed, (OR=0.47, 95% CI= 0.31 - 0.71, p< 0.001). Conclusions: In this sample of PWH drawn from an integrated health care system in California, findings suggest that enrollment via the Exchange and higher deductibles were negatively associated with some aspects of service utilization but did not impact HIV outcomes; support from ADAP appeared to help patients achieve viral suppression; ethnic disparities remain important to address in post-ACA health services for PWH even among those with access to insurance coverage.


Introduction
Major components of the Affordable Care Act (ACA) [1] were introduced in 2014 to increase access to health insurance coverage in the U.S., particularly for vulnerable populations including people with HIV (PWH). It was expected that mandates of the ACA (e.g., establishment of state insurance exchanges, and inclusion of psychiatric and substance use treatment as essential benefits) implemented in 2014 would increase health care utilization and quality, leading to improved behavioral health and HIV clinical outcomes [2][3][4].
Early evaluations suggest that some of these expectations have been met. A large multi-state study by the HIV Research Network found a decrease in Ryan White / uncompensated care in Medicaid expansion states post-ACA and a small increase in private coverage in non-expansion states [5].
Among PWH enrolled in a California health care system after ACA implementation, more new enrollees with HIV also had a primary care visit within six months of enrollment in 2014 compared with a cohort that enrolled in 2012, and rates of HIV viral control improved [6]. Research in Virginia has found improvement in viral suppression following increased access to coverage through the ACA [7,8]. Similarly, a study based on Nebraska found that insurance enrollment was independently associated with improved health outcomes including viral suppression [9].
However, little is known about the degree to which newly enrolled PWH maintain access to services within healthcare systems and what their HIV clinical outcomes are over time. In addition, specific facets of the ACA, such as enrollment via new state insurance exchanges and increased patient cost sharing via higher deductibles, have the potential to influence care. For example, a previous evaluation of PWH found that new enrollees post-ACA (2014) had more substance use disorders and were more likely to be enrolled in high-deductible plans compared to pre-ACA enrollees (2012) [6], indicating that newly enrolled PWH post-ACA may have greater behavioral healthcare needs as well as financial obstacles to accessing services.
It remains unknown to what extent enrollment through state insurance exchanges and higher deductibles influence outcomes. Exchanges provide access to tax credits, a range of coverage levels, HIV-specific guidance on planning care continuity, and information about medication benefits relevant to PWH that might not be easily accessible through other sources of coverage, e.g., through employers [10,11], and thus could have a potentially positive impact on subsequent utilization and outcomes. Deductibles may pose a barrier to accessing care, as studies in other populations have shown [12,13].
The AIDS Drug Assistance Program (ADAP) continues to play a role in maintaining access to care, providing medications, premium and out-of-pocket financial assistance to low-income PWH [14,15]. A prior study by McManus and colleagues [8] found that Virginia ADAP client enrollment in qualified health plans increased in 2015 compared with the prior year and varied based on demographic and health care delivery factors. Given the potential positive impact of ADAP on access to care even in an insured patient population, we included ADAP in our models along with enrollment via the Exchange and deductible level.
The current study investigated health care utilization and HIV control in a large health system, Kaiser Permanente Northern California (KPNC). This system has an integrated model that is becoming increasingly common [16,17], and follows the standards for health services required of plans offered on the California Exchange [18]. We anticipated that utilization would be greatest immediately after enrollment, reflecting potential pent-up demand for care. We examined factors associated with utilization as conceptualized by the Andersen model of health care utilization [19,20]. The model proposes that utilization is determined by predisposing (e.g., race/ethnicity and other demographic factors), need (e.g., HIV diagnoses) and enabling factors (e.g., benefit plan, enrollment through the Exchange) [21]. Controlling for demographic factors and ADAP participation, we expected that utilization would be higher and that HIV outcomes would be better among PWH with lower deductibles, and that enrollment via the Exchange also might have a positive effect on these outcomes.

Methods Setting
KPNC is a large, integrated health care system with over 4 million members. Members receive coverage through employers, government programs (e.g., Medicaid and Medicare) and individual plans. KPNC offers comprehensive HIV care. Treatment is provided by HIV specialists integrated into primary care, with support provided by HIV specialty nurses, case managers, and clinical pharmacists.
Patients enrolled in ADAP can fill prescriptions directly through KPNC pharmacies. Psychiatry and substance use treatment are available to members [22], as well as emergency department (ED) and inpatient care.

Measures
We obtained patient characteristics and services use from the electronic health record (EHR), and HIV clinical outcomes data from the HIV registry. Demographic variables included sex, age, and race/ethnicity. Coverage mechanism included enrollment via the California Exchange vs. other mechanisms (e.g., employer-based large group purchasers or individual plans not purchased on the Exchange). KPNC pharmacy databases include codes that indicate whether the fill was subsidized by ADAP; enrollment was assessed from HIV prescription fills in 2014. Deductible limits were classified into 3 levels (none, $1-$999 and >= $1000), as in prior studies within KPNC [23,24]. Since these may change over time, for analyses we used the values that were applicable at enrollment and every six months thereafter.
Utilization measures were aggregated in six-month intervals beginning from the post-intake date for up to 36 months (through 12/31/2017) yielding a maximum of six repeated measures. Total outpatient visit count as well as type (primary care, psychiatry, specialty substance use treatment, ED, inpatient) were summarized for the six time periods.
For HIV clinical outcomes, we created an indicator variable for each six-month period for HIV RNA suppression, defined as HIV RNA levels < 75 copies/mL [25]; on the few occasions when an individual had more than one lab result per time-period, we used the most recent observation. We used continuous measures of CD4 counts for each six-month period, using mean CD4 count in case of multiple measures per six-month period.

Analyses
Over thirty-six months post-enrollment, we examined trends in service utilization including visits to primary care (which includes HIV services), psychiatry, substance use treatment, inpatient hospitalization and ED use; and HIV clinical outcomes using bivariate statistics. To account for correlation between repeated measures, we used generalized estimating equations methodology [26]. Using multivariable logistic regression, we examined whether deductible level, enrollment via the Exchange, ADAP and demographic characteristics were associated with use of health care and HIV outcomes in the three years post-enrollment.
We modeled psychiatry and specialty substance use clinic visit counts using the Poisson distribution since we had count data (number of days) with over dispersion of zeroes (no visits) and decreasing probability of having multiple visits. We used the software's built-in capability to correct for overdispersion (the dispersion parameter is estimated by the ratio of the deviance to its degrees of freedom). The parameter estimates are not affected but the estimated covariance matrix is inflated by this factor; this is the conventional approach in Poisson regression. We included member months as an offset term to account for varying exposure length (i.e., varying length of membership due to attrition). We used the negative binomial distribution for examining primary care visits based on preliminary analyses of the distribution of visits in these categories. The exponent of the coefficient represents the rate ratio of the utilization measures over time relative to the first six-month postenrollment period.
The sample had too few Medicaid enrollees to independently examine the effects of Medicaid on study outcomes. However, we conducted a sensitivity analysis, with participants who enrolled in the health system via Medicaid removed, to determine if this changed the results of service utilization analyses. All analyses were conducted using SAS v9.4. Significance level was set at 0.05 for all analyses.

Sample characteristics
The sample included 880 PWH enrolled in the health care system in 2014 (Table 1)    We also examined the number of visits to primary care, psychiatry and substance use departments.
There was a decreasing trend in frequency over 36 months (Table 3). The most consistent and significant decreases were in primary care, which declined by 42% (1-0.58) in the six to twelve months post-enrollment period relative to the first six months after enrollment; this trend continued and by three years, participants had 53% fewer visits relative to 0-6 months post-enrollment. Those with high deductibles had 21% fewer primary care visits (RR = .79, 95% CI = 0.71-0.88, p < 0.001).  Table 3  Visits to substance use treatment also declined, but the decreases were larger after 18 months, and by three years PWH had 81% fewer substance use treatment visits. Benefit-related factors (enrollment through the Exchange, deductible levels and ADAP) were not significantly associated with visits ( Table 3).

Multivariate Analyses of HIV Clinical Outcomes
Mean CD4 counts showed a steady increase over time, with coefficients (relative to 0-6 months) increasing from 29.1 in the 6-12 months post-enrollment to 93.6 by 30-36 months (Table 4). All individuals were increasingly likely to have viral control (HIV RNA < 75 copies/mL) over time.
Approximately twice as many individuals (OR = 2.04, 1.46-2.85, p < .001) were likely to have HIV RNA < 75 copies/mL in the 6-12 twelve months post-enrollment period compared to the first six months after enrollment; they were three times as likely (OR = 3.28, 95% CI = 2.01-5.34, p < .001) to have HIV RNA < 75 copies/mL by 30-36 months post-enrollment. Enrollment through the Exchange and deductible level were not significantly associated with CD4 count or HIV RNA but enrollees in ADAP were more than twice as likely to have HIV RNA < 75 copies/mL compared to others (OR = 2.23, 95% CI = 1.35-3.68) p < 0.001). Non-white PWH were less likely to be virally suppressed, (OR = 0.47, 95% CI = 0.31-0.71, p < 0.001). --INSERT Table 4 HERE--

Sensitivity Analysis with Medicaid Enrollees Removed
The utilization models were repeated after excluding those with Medicaid, with a few minor changes noted: In analysis of predictors of number of primary care visits, the effect of ADAP enrollment became significant (RR = 1.16 (95% CI = 1.03,1.31, p = .01), rather than non-significant (p = .11, Table 3). In analysis of predictors of number of psychiatry visits, high deductible became nonsignificant (RR = 0.54, 95%CI = 0.26, 1.13, p = .10) rather than borderline significant (p = .05, Table 3). All other coefficients in all the other utilization models were substantively the same (i.e., significance or insignificance of the coefficients remained the same).

Discussion
The ACA remains an essential mechanism for increasing access to health insurance for PWH in the U.S, and it is important to examine its role in health service utilization and HIV clinical outcomes over time in combination with other factors that have an impact on PWH, including deductibles and ADAP.
Overall, our study findings indicated that use of health care was highest immediately after enrollment; and that coverage through the California Exchange and deductible level had some impact on service utilization, although the effect varied by service. ADAP benefits were associated with access to psychiatry and better HIV viral control.
We found that most of the study sample (91.8%) had a primary care visit within six months of enrollment. Since primary care in this health system included HIV treatment, this is an encouraging indicator that KPNC staffing and services were adequate overall for PWH soon after implementation of the substantial health policy changes and enrollment increases associated with the ACA. Consistent with prior studies examining utilization [27,28], we found that primary care utilization was highest immediately following enrollment and then decreased. Apart from deductibles, a factor contributing to having fewer primary care visits over time may be improvement in antiretroviral effectiveness [29,30], which has led to a decrease in the recommended frequency of laboratory testing among individuals with long-term stable viral suppression.
We examined the effects of enrollment through the Exchange and deductible levels because these ACA features continue to be a major focus of policy in the U.S. Efforts were made on the part of some California health systems [6] and the Exchange itself [10] to educate new members (and potential members) on tiers of coverage and health care initiation processes. Although it is not known if other states made similar efforts, the Exchange in California represented a novel mechanism of coverage and entailed potential challenges both in determining how to choose coverage and sign up (preenrollment) and in understanding benefits and services (post-enrollment) [4]. Our findings that enrollment through the Exchange made no difference in accessing primary care or in HIV outcomes for newly enrolled PWH suggests that regardless of mechanism, new enrollees were able to access core services and achieve viral suppression.
One recent study in California found that those with employer-based insurance had greater access to providers than those with either on-exchange and off-exchange individual private insurance plans or Medicaid [31]. Consistent with our psychiatry utilization findings, this study found worse access to primary care among those with private coverage purchased on exchanges compared to private coverage purchased individually [31]. Another study found that patients often felt overwhelmed by the array of choices offered on the exchanges and were confused by terminology and websites [32].
The reason for the effects of enrollment through the Exchange on psychiatry utilization in our sample could also be due to higher cost-sharing in Exchange plans (apart from deductibles) or to unexamined group differences such as financial constraints or fewer mental health problems among Exchange vs.
It is important to for patients and providers to understand how coverage policies affect services for PWH. One early study conducted in 2013 indicated that most PWH felt that they were not informed enough to make ACA-related decisions about their insurance [33]. HIV treatment provider knowledge regarding the ACA also is variable [34]. Our finding that higher deductibles did not impact HIV outcomes was a welcome result in light of the potential for higher deductibles to negatively impact care [35]. This may indicate that PWH were motivated to initiate primary care regardless of deductible level. ADAP financial support could also play a role in offsetting the effects of higher deductibles on access to care [36]. Prior work in the KPNC health care system found that HIV care coordinators make an active effort to "onboard" newly enrolled PWH, including ADAP enrollment [6], which could have had a positive impact on management of costs associated with deductibles as well as linkage to psychiatry. It is not known if other health systems made such efforts. However, similar to our findings, one prior study found that increased out-of-pocket spending on antiretroviral therapy associated with Medicare Part D enrollment did not have an impact on HIV viral suppression [36].
These results suggest that PWH who face higher deductible costs, at least in some health systems, have been able to manage these obligations without compromising viral suppression.
It is worth noting that we found worse HIV outcomes for non-White PWH, despite similar utilization of primary care, mental health, and substance use treatment. Race/ethnic disparities in HIV care are a longstanding concern in the HIV treatment field. It was hoped that disparities would be at least partially mitigated post-ACA [4,37], although recent data indicate that non-whites continue to have worse HIV care outcomes across multiple health care settings [38][39][40]. It is possible that factors associated with race/ethnicity and worse HIV outcomes but not measured in our study, such as health literacy [41] and variability in use of electronic provider communication tools [42,43] could have contributed to these differences. The race/ethnic disparities in HIV clinical outcomes observed in our sample, in which overall insurance coverage was not a barrier to care, highlight the importance of addressing this ongoing challenge to health equity.

Study Strengths And Limitations
The study was conducted in a large integrated health system with access to data on type of insurance coverage, enrollment mechanisms, ADAP, use of health services and routine laboratory measures; and measured outcomes over three years following ACA implementation. However, limitations should be noted. The California Exchange continues to modify its coverage options [44,45], and the effects of enrollment via the Exchange on outcomes of interest are likely to shift over time. Data on insurance coverage and viral control prior to KPNC enrollment were not available. Loss to follow up is also a limitation: although retention was high, HIV outcomes were based on PWH who remained in care and could be high compared to those who did not complete routine laboratory testing or left the health plan during study follow-up.
Although Medicaid expansion has benefitted PWH [46,47], too few Medicaid beneficiaries were identified to examine separately. Yet repeating our analyses with Medicaid beneficiaries excluded, as a sensitivity analysis, resulted in few changes to the results. There is variability in the ways that patients can use ADAP in California (i.e., support for medication purchases as well as help in purchasing insurance and covering co-pays) [48], which we were not able to examine. However, we included ADAP status using prescription records as an indicator of whether study participants had received financial assistance through this program.
The impact of the ACA on HIV care may vary by state [49], limiting generalizability. The sample was drawn from a single institution in California and participants had relatively high levels of viral control relative to PWH in other settings such as Ryan White clinics [40,50]. However, economic barriers to service utilization for PWH with access to care are of concern in many health systems, including systems in other states and countries, and our study contributes to this important area of inquiry. In addition, conducting the study in single large integrated health care allowed us to examine ACA-related research questions without needing to control for variability across health care insurance and treatment providers.

Conclusions
This study examined factors associated with use of health care and HIV control among PWH following ACA implementation in a California health system. Outpatient utilization was highest immediately after enrollment and subsequently decreased. Enrollment through the Exchange and higher deductibles were negatively associated with psychiatric services, but did not appear to impact HIV outcomes, which improved over time and were associated with ADAP enrollment. Race/ethnic disparities in HIV clinical outcomes remain critical to address among PWH post-ACA, even in integrated health plans. It is essential that PWH maintain continuous access to health care in order to benefit fully from pharmacological advances in HIV treatment and to manage common medical and psychiatric comorbidities. As health care reform continues to develop, health care providers and policy leaders must consider potential effects on PWH.

Declarations
Availability of data and materials The datasets generated and/or analyzed during the current study are not publicly available. Active partnerships between external researchers and the parent study team facilitates more productive collaborations that will effectively leverage the investment to develop the data as compared with generation of public use datasets. Upon request, de-identified data will be made available as study manuscripts are published. However, there remains the possibility of deductive disclosure of participants with unusual characteristics and disclosure of Kaiser Permanente proprietary information.
Thus, researchers who seek access to individual level data will be required to sign a data sharing agreement. After the end of the project funding period, the cost of computer programmers preparing de-identified datasets suitable for sharing will be borne by the party requesting the data.

Authors' Contributions
DDS and CIC conceived the study and obtained funding. SP designed the analysis plan and completed data analysis. All authors contributed to data interpretation as well as drafting and revising the manuscript. All authors read and approved the final manuscript.